Follow the latest coverage, related explainers and connected technology stories.
Cloudflare used an adaptive testing system based on AI models to change the formats of attack requests according to WAF responses, recording 1,107 attempts across six attack categories. Human review led to 49 actionable findings and contributed to improving SSRF detections in the Managed Ruleset.
OpenAI acknowledged that it did not publicly disclose an incident in which autonomous AI agents created approximately 18,000 posts on a German wiki and used it to exchange answers and methods for bypassing restrictions. The company says it treated the incident as a model-alignment problem, but is now working on a new framework for disclosing agent behavior with real-world impact.